v, con - American Chemical Society

The low value predicted by the INDO method for the unique OCO angle (65 "), as well as the results of the electron-population analysis, strongly sugge...
0 downloads 0 Views 561KB Size
7286

Molecular Geometry and Electronic Structure of Carbon Trioxide' John F. 01sen2 and Louis Burnelle Contributionfrom the Department of Chemistry, New York University, New York, New York 10003. Receiued June 9, 1969 Abstract: The extended Hiickel and INDO methods have been applied to various molecular geometries of COS

(pyramidal, symmetric trigonal, and Y-shaped). Both methods predict the Y-shaped structure (of Czvsymmetry) to be favored in the ground state, in agreement with infrared observations. The electronic ground state of the molecule is 'Al, correlating with 'E' in Dah symmetry, so that the adoption of the Czvsymmetry can be related to the Jahn-Teller effect. The low value predicted by the INDO method for the unique OCO angle (65 "), as well as the results of the electron-populationanalysis, strongly suggest a closed-ring structure for the compound, with a carbonyl bond. arbon trioxide, COS,has recently been isolated and identified by Moll, Clutter, and Thompson, using matrix isolation technique. This unusual oxide of carbon was first postulated by Katakis and Taube4 as a kinetic intermediate in the reaction of oxygen atoms and CO,. Carbon trioxide is, to our knowledge, the first reported 22-valence electron AB3-type molecule. AB3 molecules containing 24, 25, or 26 valence electrons are of course well known,s e.g. NO,-, C032-, C103, and IO3-. These AB, molecules have planar or pyramidal structures in their ground electronic state^.^ At present, only spectrally detectable quantities of cos have been prepared and the most important method of preparation of the compound is probably the photolysis of solid CO, at 77°K using a xenon resonance lamp.6 The most probable reactions leading to the production of CO, are6

C a

COzhv_ COz* +CO

COz

+ 0 ('D)

+ 0 (ID) +COa

where C02* represents an electronically excited COS molecule. Moll, et al.,, have considered structures I-VI as reasonable for a species resulting from the reaction of an oxygen atom with a C 0 2 molecule. However, the anal-

111, C,"

0/c\o

'O/ IV,czv

od~0-0

v,con

o-c-o-@ VI, cs

ysis3 of the infrared spectral data of normal and isotopically substituted carbon trioxide was largely in favor of a planar Clv molecule, namely 111. (1) Work supported by the U. S. Army Research Office (Durham). (2) Address correspondence to: Department of Chemistry, Staten Island Communitv Colleee. Staten Island. N. Y. 10301. (3) N. G . Moll: D. R.kiutter, and W. 'E. Thompson, J . Chem. Phys., 45. 4469 (1966). (4) D. Katakis and H. Taube, ibid., 36, 416 (1962). (5) A. D. Walsh, J. Chem. Soc., 2301 (1953). (6) K. V. Krishnamurty, J . Chem. Educ., 44, 594 (1967).

We wish to present herein the results of some extended Hiicke17 (XHMO) and some INDOE molecular orbital calculations on C 0 3 . The computational details have been presented e l ~ e w h e r e and ~ , ~ will not be given herein. The calculations have been carried put by assuming the three bond lengths equal to 1.20 A throughout.

Results and Discussion The major part of this study is focused primarily on the D 3 h , C3v,and Cpvsymmetry species of COO. All of these species are related via an angular displacement of the internal coordinates and they presumably result from the reaction of a singlet oxygen atom with a CO, molecule at the carbon center of the The other species that are possible for COS (IV-VI) apparently result from the interaction of a singlet oxygen with *~ an oxygen center of the CO, m ~ l e c u l e . ~Extensive bond-length minimization procedures would be necessary to find the most stable structure for these latter species. We have elected not to undertake in this work any absolute minimization, especially in view of the fact that the computational methods used are not to be trusted, in general, for their bond-length i n f o r m a t i ~ n . ~ ~ ~ The computational methods d o generate accurate information when only bond-angle variations are considered, howe~er.~!~O Consequently, we have assumed as a first approximation that the D3hspecies of COOcan be related to the CSvand the CZvspecies by means of bond-angle variatoions only, leaving all CO bond lengths fixed at 1.20 A. The XHMO calculations predict the twohighest energy electrons of Dah C 0 3 to reside in a degenerate molecular orbital, namely 3e'. According t o the extended Hiickel method, a distortion to a pyramidal CaVspecies is not predicted to occur. Thus, when the bond angle is 105" the CaVstructure lies 2.14 eV higher in energy than the D a h configuration; for a bond angle of 90" it lies 7.03 eV higher than the latter. On the other hand, as will be seen shortly, the XHMO method predicts that an in-plane distortion of COe leading t o a Czvsymmetry species is very favorable. We illustrate in Figure 1 the XHMO orbital-energy correlation diagram for COI. It gives the orbital energies us. an in-plane OCO angular distortion. (7) J. F. Olsen and L. Burnelle, J . Phys. Chem., 73,2298 (1969). (8) J. A. Pople, D. L. Beveridge, and P. A. Dobosh, J. Chem. Phys., 47, 2026 (1967). (9) L. C. Cusachs and B. B. Cusachs, J . Phys. Chem., 71, 1060 (1967). (10) J. A. Pople, D. L. Beveridge, and N. S. Ostlund, Infern. J . Quant. Chem., 1 , 293 (1967).

Journal of the American Chemical Society / 91:26 1 December 17, 1969

7287 ,Cor;,4,b21'

1 I

13.

14.

s

-15. -

2 2,

P W C

1 I

15.

1

21.

22.

23

-524

~ i k i l i o

iio

ti0

iio

ibo

ab

$0

jo

io

U (degrees)

-2''1500

130"

110"

Q

(degrees)

Figure 2. Energies of the lower configurations of in-plane bending.

90'

Figure 1. XHMO orbital energy diagram of COa,as a function of the unique OCO angle a.

(The abscissa is the unique OCO angle, which is labeled a.) It is essentially a Walsh5 correlation diagram for AB, molecules, although Walsh in constructing diagrams for AB, systems considered only the out-of-plane distortion leading to a pyramidal structure. For clarity, the five lowest and the four highest energy molecular orbitals have been omitted from Figure 1. Of the two components of the degenerate 3e' molecular orbital, the 4b2level is seen to rise sharply as a decreases from 120", due to the repulsive interaction with the 3b2 molecular orbital (this is an application of the noncrossing rule). The other component, namely 5al, is seen, on the other hand, t o sharply stabilize at low values of a. It is primarily this stabilization of the 5al molecular orbital that accounts for the low calculated unique equilibrium angle in C o t (vide infra). The configuration (3e')2 gives rise to three electronic states, namely ,A2', lE', and lA1'. They most probably lie in the order given here, the triplet being the lowest. Since the extended Hiickel method does not take the interelectronic repulsion explicitly into account, it cannot make a distinction between the three states, which are predicted to coincide. As the molecule distorts t o Czv symmetry, the splitting of the 3e' orbital enables the extended Hiickel method to distinguish three states, corresponding respectively to the three configurations (5a1)2,(5a1)(4bz), and (4bz)2(outside a core). Their energies us. the unique OCO angle a are plotted in Figure 2. The energy of the (5a1)2 configuration exhibits a very deep minimum in the vicinity of a = 82",

Of

CO3 as a function

that of (4b2)2a shallow minimum around 135". The state associated with (5a1)(4bz) has a shallow minimum near 130". As a matter of fact, this is actually a mixture of singlet and triplet states and in the case where the splitting of the degenerate orbital is small, it may happen that the triplet is actually the ground state. However, in COI the splitting is remarkably large (4.95 eV) at the equilibrium angle predicted for the (5a1)2 configuration, so that on the basis of the extended Hiickel results one can predict that the ground state of COSis 'AI (correlating with 'E' in DBh symmetry). Its electronic configuration is predicted to be (la1)2(2a1)2( 1bz)2(3a1)2( 1bl)2(4al)2(2b2)2(2bl)2( 5al)2(3 1a# : lA1. For values of a larger than 120", the splitting of the 3e' orbital is considerably smaller. At the angle minimizing the energy of the (4b2)(5al) configuration, the splitting amounts to 0.15 eV only. Under those conditions it is most probable that the triplet is actually the lowest state for these values of the angle. In a study on methylenes, Hoffmann has proposed a criterion according to which the triplet state is the lowest when the splitting is less than 1.5 eV. Even if we do not know the form the criterion should take in CO,, in the case under discussion the splitting is so small that there is little doubt that our conclusion is valid. According t o Hund's rule, the triplet must be the lowest state for the symmetric trigonal arrangement of the molecule. One may thus deduce that the curve of the triplet (,B2, correlating with 3A2'in D3h symmetry) intersects the curve of the ground state (lA1) for a smaller than 120" and remains the lowest state for larger values of a. (11) R. Hoffmann, G. D. Zeiss, and G. W. Van Dine, J . Amer. Chem. Soc., 90, 1485 (1968); R. Hoffmann, ibid., 90, 1475 (1968).

Olsen, Burnelle

Molecular Geometry and Electronic Structure of COS

7288

a>

IZOO

a = 1200

a